6 th annual international symposium on advanced radio technologies

6thANNUALINTERNATIONAL SYMPOSIUM ON ADVANCED RADIO TECHNOLOGIES

Rapidly Deployable Broadband Communications for Disaster Response

Charles W. Bostian, Scott F. Midkiff, Timothy M. Gallagher, Christian J. Rieser, and Thomas W. Rondeau

Center for Wireless Telecommunications

Virginia Tech

Blacksburg, VA 24061-0111

March 4, 2003

Testbed for High-Speed ‘End-to-End’ Communications in Support of

Comprehensive Emergency Management

A Project of

Virginia Tech’s Center for Wireless Telecommunications (CWT)

Science Applications International Corporation

Sponsored by the National Science Foundation(Digital Government Program Award 9983463)

With additional support from the National Response Center (NRC)C.W. Bostian, PI, S.F. Midkiff, and W. M. Kurgan

Co-PIs

Disaster Communications Project Research Objectives

Overall Objective: Develop a rapidly deployable high capacity backbone radio system that seamlessly links on-site communications systems and networks to each other and to the outside world.

Technical Challenges: find short-lived radio paths of opportunity and compensate for the shortcomings of these paths at both the radio and the network levels to deliver optimum performance.

The Situation Our Project Addresses – A disaster area several miles on a side in which almost all communications have been wiped out.

Cell phones provide voice service but are quickly saturated.

Landline connections exist at the perimeter of disaster area that could be accessed via high-speed wireless links.

Responders need rapidly deployable broadband communications. On September 11, getting needed connectivity took about a week.

Fiber or Wireless NetworkConnection

2-3 km range(to be extended)

Hub

Hub “illuminates” disasterarea with 120 Mbpsconnectivity using broadband wireless (originally 28 GHz commercial LMDS equipment).

Our Project: System Concept

120 Mbps

RemoteUnit

120 Mbps

Disaster Area

RemoteUnit

High Data Rate Cable/Wireless Connection

LMDS

IEEE802.11b

Hub

Remote GISand other services

10/100BaseT

RemoteRemote

“Virtual Ethernet”

RemoteLAN

RemoteLAN

Existing NetworkInfrastructure

System Overview

VoIP (non-cellular) Wireless

TelephonesVoIP

Wireless Telephones

Why did we use the 28 GHz LMDS band?

Remember wireless data technology in 1998!

• IEEE 802.11b was an 11 Mbps indoor LAN for the office market.

• Only one or two companies offered radio modems that worked above 100 Mbps. These were intended for satellite links and priced accordingly.

• Successful 155 Mbps transmission over a wireless link was an expensive laboratory demonstration.

• The Dot.Com Bubble was on and equipment manufacturers were selling everything they could possibly manufacture. They were not very interested in the disaster relief market.

With the IEEE 802.11x and 802.16 technology now available, we are in the process of changing our operating frequency to the 5 GHz UNII band and using Proxim Tsunami Radios. But we learned a lot about 28 GHz.

28 GHz works best with LOS but NLOS paths exist and will work. These “paths of opportunity”can come from any direction and differ significantly from each other in radio channel characteristics.

Researchers disagree about their characteristics

To study the broadband characteristics of 28 GHz bounce paths, we developed an impulse channel sounder using UWB technology.

28 GHz paths of opportunity involve rough surface scattering and produce a continuous distribution of multipath components.

Power delay profiles for scattering by limestone and brick walls.The first waveform is the LOS signal and the second is the scattered signal. From C. L. Dillard, T. M. Gallagher, C. W. Bostian, D. G. Sweeney. “28 GHz scattering by brick and limestone walls,” 2003 IEEE Antennas and Propagation Society International Symposium, Vol. 3, pp. 1024-1027

5 10 15 30 45

020406080

100

Excess delay [ns]

Angle [degree]

-15 dB Maximum Excess Delay (d1=d2=9m)

Limestone

Brick

Limestone 85 77.5 76 45 17

Brick 18 15 12 13 18

5 10 15 30 45

Because of rough surface scattering, paths of opportunity can have significantly different channel characteristics.

Calculated BER for Burruss path

Calculated BER for Cowgill path

Measurement Layout

Calculated plots of BER versus

Eb/N0 for paths whose power

delay profiles we measured.

We need a radio that is capable of finding these paths of opportunity and configuring itself to make optimal use of them.

In addition, there is a human factors problem.

Fire fighters and other first responders emphatically do not want a radio that requires hands-on operation or frequent adjustment by an expert. They need a radio that is smart enough to find the best path of opportunity, configure itself for it, and close a link, all without human intervention.

In other words, they need a cognitive radio.

Adaptive radios can adjust themselves to accommodate anticipated channels and environment

Fixed radios are set by their operators

Cognitive radios can sense their environment and learn how to adapt

A cognitive radio is not necessarily a software radio!

Beyond adaptive radios, cognitive radios can handle unanticipated channels and environments

Cognitive radios require:• Sensing• Adaptation• Learning

Cognitive radios can maintain a quality of service (smart transmitter) and reduce interference to neighboring radios (smart receiver)

What is a Cognitive Radio?

How can we model a cognitive radio?

• Synthesize biological cognition model and genetic algorithms (GA)– Fundamental principles evolve to optimize

response through genetic crossover

– Random discoveries and spontaneous inspiration occur through genetic mutation

• Biological model of cognition– Left brain processing – “Logical”– Right brain processing – “Creative”

resourcesEnvironment

resourcesRadio

Adaptation and Processing

- Learning

Monitor and Control

- Exploit trends

Our Approach to Realizing a Cognitive Radio

“Controlled Chaos” Allow evolution through unique solutions Force compliance with regulations

Use biologically-inspired genetic algorithms to control and adapt the radios

System is designed in terms of chromosomes

Represents radio parametersChannel model also encapsulated in a chromosome

Like biological organisms, the radio will play and learn from current and past behavior

DNA Strand, Gilbert – Apple, 2003 Performing Feats of Bioinformagic

http://www.apple.com/pro/science/gilbert/

Cognitive Radio System Diagram

Channel Model(HMM)

“Take action to adapt radio”

“How to play: Use channel models and radio parameters to enable radio evolution”

Radio HardwareChannel

Estimation Process

Radio Baseband processor

(PHY and MAC layers)

Evaluation functions, child chromosomes, tags/templates

Radio Parameters / Performance Statistics

Wireless Channel Genetic Algorithm

(WCGA)

Wireless System Genetic Algorithm

(WSGA)

System Monitor Chromosome

Cognitive System Monitor (CSM) (Learning Classifier, Meta-GA, Memory, Control)

“Model channel”

Channel (Data and RF)

WCGA and HMMs for Wireless Channels

The WCGA uses CWT’s broadband sounder technology to observe the channel.

Wireless Channel Genetic Algorithm

(WCGA)

CWT’s Broadband Channel Sounder

The channel is then modeled by a Hidden Markov Model (HMM) that is compact and computationally efficient.

WCGA and HMMs for Wireless Channels

Hidden Markov Model

0

100

200

300

400

500

600

700

800

900

1000

1100

1200

1300

0 1 2 3 4 5

Burst Error Length

Fre

qu

en

cy o

f B

urs

t E

rro

rs WCGA

Simulated

Radio Adaptation – Improved performance over time

Channel Impulse Response

Burst Error Length

Fre

qu

ency

of

Bu

rst

Err

ors

Am

pli

tud

e (V

)

Time (ns)

CSM and WSGA for Intelligent Radio Adaptation

Wireless System Genetic Algorithm

(WSGA)

Cognitive System Monitor (CSM) (Learning Classifier, Meta-GA,

Memory, Control)

The WSGA takes the channel model and information from the CSM to adapt the radio using a Genetic Algorithm.

The CSM synthesizes channel model and adaptation process and better directs radio evolution from observed and learned behavior.

Current Work – Proxim Tsunami Radios

Adapting with limited range of adaptable parameters:

Modulation: QPSK, QAM8, QAM16 Power: 6 dBm – 17 dBm Frequency: See figure on left Uplink/Downlink ratio

Even with this limited-flexibility legacy radio, we can use our cognitive processes to adapt the radio, including the avoidance of an interferer.

Frequency Channels available to Proxim Tsunamis

Interference Test setup

Network Base Station

UnitNetwork

Subscriber Unit

Interfering Unit

Cognitive RadioAlgorithms

PC

Video Demo (Mac)

Wireless Router

Proxim BSU Proxim SUWireless

Router

Cognitive RadioAlgorithms

PC

VideoDemo (Mac)

PC control Proxim BSU

TelemedicineDemo

Wireless Laptop

192.168.0.13

192.168.1.113 192.168.0.1

192.168.0.100

192.168.1.16 192.168.1.100

192.168.0.101

192.168.0.2

192.168.0.12

192.168.0.112

192.168.0.?

192.168.0.?

Connection to general IP internet802.11b/g

802.11b/g

802.11b/g

Interferer(INT)

Cognitive Radio LinkNetwork under test (NUT)

20/30/40/60 MB/sec

5 Ghz unlicensed

5 Ghz unlicensed

10/100BaseTEthernet

10/100BaseTEthernet

10/100BaseTEthernet

10/100BaseTEthernet

10/100BaseTEthernet

10/100BaseTEthernet

Module A Module B

Module C

Demo Physical Layout

B

A C

NUT

INT

Cognitive Radio Testbed Demo

• Cognitive Radio Testbed Link

Cognitive Radio Testbed Demo

• Interferer Degrades Broadband Wireless Link

• WSGA evolves radio operation

• Link quality of service (QOS) is restored even in presence of interferer

Interferer Broadband Wireless Link

InterfererTx=11dBmQAM 16Fibs 123/4 FECFreq=b-5

LinkTx=6dBmQAM 16Fibs 223/4 FECFreq=a-5

LinkTx=16dBmQPSK 4Fibs 41/2 FECFreq=a-5

Future Work: Building A More Complete Cognitive Radio

Build an adaptive radio88 MHz – 6 GHz

100 Mbps data link

Programmable power and diversity techniques

Arbitrary signal constellations for increased modulation capability

Programmable MAC layer for timing, payload sizes, and connection maintenance

Extend cognitive radio algorithmsInclude MAC layer adaptation using economic market analysis models

Improve sensing, modeling, and learning

Future Work: Cognitive Radio Network

• Extend cognitive radio techniques to build network test bed

Analyze cognitive radio network coordination via economic theory

• Investigate adaptation mechanisms at MAC layer

Apply genetic algorithm approach to MAC/Data Link Layer

Provide a requested quality of service

Enable dynamic spectrum accessGene Splicing Presentation

Biotechnology Onlinehttp://www.biotechnology.gov.au/biotechnologyOnline/interactives/

gene_splicing_interactive.htm

References• C.W. Bostian, S. F. Midkiff, W. M. Kurgan, L. W. Carstensen, D. G. Sweeney, and T. Gallagher, “Broadband Communications for Disaster Response”, Space Communications, Vol. 18 No. 3-4 (double issue), pp. 167-177, 2002

• C.L. Dillard, T.M. Gallagher, C.W. Bostian, and D.G. Sweeney, “Rough Surface Scattering From Exterior Walls at 28 GHz,” IEEE Trans. Antennas and Propagation, in press.

• T.W. Rondeau, C.J. Rieser, T.M. Gallagher, and C.W. Bostian. “Online Modeling of Wireless Channels with Hidden Markov Models and Channel Impulse Responses for Cognitive Radios,” 2004 International Microwave Symposium

Contact InformationCharles W. Bostian bostian@vt.edu

Scott F. Midkiff midkiff@vt.edu

Timothy M. Gallagher tgallagh@vt.edu

Christian J. Rieser cjrieser@vt.edu

Thomas W. Rondeau trondeau@vt.edu

http://www.cwt.vt.edu

540-231-5096This work was supported by the National Science Foundation under awards 9983463 and DGE-9987586.